Robust Bayesian inference in proxy SVARs

We develop methods for robust Bayesian inference in structural vector autoregressions (SVARs) where the parameters of interest are set-identified using external instruments, or ‘proxy SVARs’. Set-identification these models typically occurs when there multiple instruments shocks. Existing approaches to proxy SVARs require researchers specify a single prior over model’s parameters, but, under set-identification, component is never revised. extend approach proposed by Giacomini and Kitagawa press[a] – which allows relax potentially controversial point-identifying restrictions without having an unrevisable SVARs. provide new results on frequentist validity also explore effect instrument strength about identified set. illustrate our revisiting Mertens Ravn (2013) relaxing assumption that they impose obtain point identification.